Nature - USA (2019-07-18)

reSeArCH Letter

WNT signalling pathways (Extended Data Fig. 2i, k), and had higher
levels of expression of FOXA1 mRNA relative to the wild-type cases
(Extended Data Fig. 2l). Together, these data suggest that class-1 muta-
tions emerge in localized prostate cancer, whereas class-2 and class-3
mutations are acquired or enriched, respectively, in the course of
disease progression.
Class-1 mutations consist of missense and in-frame indels that clus-
ter at the C-terminal edge of the winged-helix DNA-binding FKHD.
The majority of the class-1 mutations were located either within
the wing-2 region (residues 247–269) or a 3D hotspot that spatially
protrudes towards wing  221 (Fig. 2a, b, Extended Data Fig. 3a, b).
Notably, these mutations did not alter FKHD residues that make base-
specific interactions with the DNA^22 ,^23 (Fig. 2a, Extended Data Fig. 3c).
In FOXA proteins, wing-2 residues make base-independent (that is,
non-specific) contacts with the DNA backbone^23 ,^24 that reportedly
impede its nuclear movement^24. Thus, we hypothesized that class-1
mutants with altered wing-2 regions would display faster nuclear
mobility.
We cloned representative class-1 mutants of FOXA1: I176M
(mutation of the 3D hotspot), R261G (missense) and R265–Q271del
(in-frame deletion), all of which retained nuclear localization
(Extended Data Fig. 3d). In fluorescence recovery after photobleach-
ing (FRAP) assays, we found class-1 mutants had 5–6× faster nuclear
mobility irrespective of the mutation type (Fig. 2c, d, Extended Data

Fig. 3e, g). By contrast, class-2 mutants with intact wing-2 regions were

sluggish in their nuclear movement (Fig. 2d, Extended Data Fig. 3f, g).

Using single particle tracking, we verified that class-1 mutants have a

higher overall rate of nuclear diffusion, with 3–4-fold fewer slow par-

ticles and shorter chromatin dwell times (Extended Data Fig. 3h, i).

In chromatin immunoprecipitation with parallel DNA sequencing

(ChIP–seq) assays, we found that ectopically expressed class-1 mutants

in HEK293 cells bind DNA at the consensus FOXA1 motif (Extended

Data Fig. 3j, k). In prostate cancer cells, the class-1 cistrome entirely

overlapped with wild-type binding sites, with similar enrichment

for FOXA1 and AR cofactor motifs, AR-binding sites and genomic

distribution (Extended Data Fig. 3l–s). Furthermore, in growth rescue

experiments using untranslated-region-specific small interfering (si)

RNAs that targeted the endogenous FOXA1 transcript, we found that

exogenous class-1 mutants fully compensated for the wild-type protein

(Extended Data Fig. 4a).

Next, we asked how class-1 mutations affect AR signalling. Similar to

wild-type FOXA1, both class-1 and class-2 mutants interacted with the

AR signalling complex (Extended Data Fig. 4b–d). In reporter assays,

class-1 mutants induced 3–6-fold higher activation of AR signalling

(Fig. 2e), which was evident even under stimulation with castrate levels

of androgen or treatment with enzalutamide (Extended Data Fig. 4e, f).

In parallel assays, class-2 mutants showed no differences relative to

wild-type FOXA1 (Fig. 2e). Transcriptomic analyses of class-1 tumours

a

P = 0.001

1

2

3

4

5

6

7

8

Class 1Class 2

Mutation frequency (%)

Variant allele frequency (%)

Primary Met.

9

0

2

4

6

8

10

Missense

In-frame

Frameshift

FOXA1functionaldomains

TD FKHD RD

100

200

300

400

Class 1 Class 2

Amino acid 1 275 472

53%

7%

13%

31%

16%

49%

29%

ETS

SPOP

FOXA1

AR

WNT

PI3K

DRD

050100150

No. of cases (total 371)

200

0

2

46

No. of mutations

Pathways

MIPOL1 TTC6

FOXA1

Chr14

Chr14

Chr15

Chr16

Chr17

Chr18

Chr19

Chr20

Chr21

Chr22

ChrX

ChrY

ETV1

MYC

Chr2Chr1

Chr4Chr3

Chr5

Chr6

Chr7

Chr8

Chr9

Chr10

Chr11

Chr12

Chr13

DP

TL

I

D

Class 3

0

10

20

30

Locus rearrangement frequency (%)GAPDH FOXA1 MYC

Benign Primary Met.

Class 3:

P = 7.90 × 10 –10

P = 0.0041 P = 0.26

Primary Met.

Class 1Class 2

20

40

60

0

10

20

Class 1 Class 2 Class 3

FOXA1 alteration

frequency (mCRPC) (%)

25

15

5

65.4%

34.6%

FOXA1 mutated/

locus rearranged WT

FOXA1

MYC

ERG

TMPRSS2

0%

10%

20%

30%

40%

50%

0 10,000 20,000 30,000 40,000 50,000

Ranked locus

Locus rearrangement frequency

Chr21q21

Chr8q24

Chr14q21

Others

0

2

4

6

8

10

0

2

4

6

8

10

b

cd

ef gh

Fig. 1 | Structural classes of FOXA1 alterations. a, FOXA1 mutations and
key alterations in mCRPC. Alterations in ETS, AR, WNT, PI3K and DNA
repair (DRD) were aggregated at the pathway or group level. b, Locus-level
recurrence of RNA-seq structural variations. c, Structural classification of
FOXA1 mutations. TD, transactivation domain; RD, regulatory domain.
d, Structural classification of FOXA1 locus rearrangements. DP, tandem
duplications; TL, translocations; I, inversions; D, deletions. e, Frequency of
FOXA1 mutational classes by prostate cancer stage (n = 888 primary,

658 metastatic (met.)) (two-sided Fisher’s exact test). f, Variant allele

frequency by stage and class (two-sided t-test). Box plot centre, median;

box, quartiles 1–3, whiskers, quartiles 1–3 ± 1.5 × interquartile range

(IQR). g, Locus-level recurrence of structural variants based on RNA-

seq by prostate cancer stage (two-sided Fisher’s exact test). h, Integrated

(RNA-seq and whole-exome sequencing) recurrence of FOXA1-alteration

classes in mCRPC (Stand Up 2 Cancer and Michigan Center for

Translational Pathology (MCTP) cohort, n = 370).

414 | NAtUre | VOL 571 | 18 JULY 2019